0013-7227/01/$03.

00/0 The Journal of Clinical Endocrinology & Metabolism 86(8):3761–3767

Printed in U.S.A. Copyright © 2001 by The Endocrine Society

TNF-␣ and Hyperandrogenism: A Clinical, Biochemical,

and Molecular Genetic Study
HÉCTOR F. ESCOBAR-MORREALE, ROSA M. CALVO, JOSÉ SANCHO, AND JOSÉ L. SAN MILLÁN

Departments of Endocrinology (H.F.E.-M., R.M.C., J.S.) and Molecular Genetics (J.L.S.M.), Hospital Ramón y Cajal, 28034
Madrid, Spain

Downloaded from [Link] by guest on 07 May 2023

To evaluate the role of TNF-␣ in the pathogenesis of hyperan- ⴚ238G/A, and ⴚ163G/A) were equally distributed in hyperan-
drogenism, we have evaluated the serum TNF-␣ levels, as well drogenic patients and controls. However, carriers of the
as several polymorphisms in the promoter region of the TNF-␣ ⴚ308A variant presented with increased basal and leuprolide-
gene, in a group of 60 hyperandrogenic patients and 27 stimulated serum androgens and 17-hydroxyprogesterone
healthy controls matched for body mass index. levels when considering patients and controls as a group. No
Hyperandrogenic patients presented with mildly increased differences were observed in serum TNF-␣ levels, body mass
serum TNF-␣ levels as compared with controls (mean[median] index, and insulin resistance indexes, depending on the pres-
ⴞ SD: 7.2[7.0] ⴞ 3.3 pg/ml vs. 5.6[4.4] ⴞ 4.0 pg/ml, P < 0.02). ence or absence of these variants.
Although no differences in body mass index and insulin re- In conclusion, our present results suggest that the TNF-␣
sistance indexes were observed between patients and con- system might contribute to the pathogenesis of hyperandro-
trols, when subjects were classified by body weight, serum genism, independent of obesity and insulin resistance. How-
TNF-␣ was increased only in lean patients as compared with ever, elucidation of the precise mechanisms underlying the
lean controls, but this difference was not statistically signif- relationship between the TNF-␣ system and androgen excess
icant when comparing obese patients with obese controls. is needed before considering TNF-␣ as a significant contrib-
The TNF-␣ gene polymorphisms studied here (ⴚ1196C/T, uting factor to the development of hyperandrogenism. (J Clin
ⴚ1125G/C, ⴚ1031T/C, ⴚ863C/A, ⴚ857C/T, ⴚ316G/A, ⴚ308G/A, Endocrinol Metab 86: 3761–3767, 2001)

H YPERANDROGENISM, OR ANDROGEN excess, is

possibly the most frequent endocrine disorder in
women of reproductive age. Hirsutism is a clinical manifes-
TNF-␣ in adipose and muscle tissue has been proposed to
play a key role in the development of insulin resistance in
humans (10, 11) by decreasing the tyrosine kinase activity of
tation of androgen excess and is present in 7.1% of Spanish the insulin receptor (12). This effect is mediated by the
women in this age range (1). The most common cause of insulin-receptor substrate-1 (12), is associated with an in-
hirsutism is polycystic ovary syndrome (PCOS), as defined creased expression of TNF␣ receptors in adipose tissue (13,
by endocrine criteria (2), showing a 6.5% prevalence in Span- 14), and can be antagonized by troglitazone in experimental
ish women (1). animals (15). Although TNF␣ acts through autocrine-para-
The increasing evidence that hyperandrogenism and crine mechanisms in adipose and muscle tissue, increased
PCOS have a genetic basis (3, 4) has stimulated research into serum levels of TNF-␣ have been found in several conditions
associated with insulin resistance, such as type 2 diabetes
the genes involved in the pathogenesis of these disorders.
mellitus (16) and obesity (17).
Despite significant efforts, the precise genetic mechanisms
TNF␣ is also a significant source of genetic variability.
leading to hyperandrogenism remain unknown, suggesting
There are several single nucleotide polymorphisms in the
that this disorder is a complex trait in terms of inheritance.
regulatory region of the TNF␣ gene, and some of them have
TNF-␣ influences the reproductive axis, inducing changes been proposed to play a role in the pathogenesis of insulin
that closely resemble those found in patients with PCOS and resistance, type 2 diabetes mellitus, and obesity (18 –20),
hyperandrogenism. TNF␣ stimulates proliferation and ste- among other disorders.
roidogenesis in rat theca cells in vitro (5, 6), facilitating the Hyperandrogenism, including PCOS and hirsutism with
effects of insulin and IGF-I in a dose-dependent and additive normal ovulation, is present in as many as 50% of women
fashion (6). Moreover, TNF␣ may be involved in apoptosis diagnosed with type 2 diabetes mellitus (21); conversely,
and anovulation in the rat ovary (7). In humans, increased impaired glucose tolerance is a frequent finding in PCOS and
serum TNF␣ levels have been found in lean PCOS patients non-PCOS hyperandrogenic patients (22, 23). The association
(8), but no differences in the follicular TNF␣ content have of obesity with hyperandrogenism is also well known (24).
been found between normal and polycystic ovaries (9). Therefore, the genes related to type 2 diabetes mellitus and
TNF␣ has several metabolic activities. Hyperexpression of obesity are considered candidate genes for the inheritance of
hyperandrogenism (25).
The present study was undertaken to further delimitate
Abbreviations: BMI, Body mass index; ⌬4-A, ⌬4-androstenedione; F,
cortisol; DHEAS, dehydroepiandrosterone-sulfate; FT, free testosterone; the influence of TNF␣ on the pathogenesis of female hy-
FOH, functional ovarian hyperandrogenism; 17-OHP, 17-hydroxypro- perandrogenism, using a clinical, biochemical, and molecu-
gesterone; PCOS, polycystic ovary syndrome. lar genetic approach.

3761
3762 The Journal of Clinical Endocrinology & Metabolism, August 2001, 86(8):3761–3767 Escobar-Morreale et al. • TNF-␣ and Hyperandrogenism

Materials and Methods levels using the homeostatic model assessment (HOMAIR) (33) and
using the quantitative insulin sensitivity check index (QUICKI) (34).
Subjects
Patients were classified clinically as follows: Women presenting with
Sixty consecutive hyperandrogenic patients, aged 24.3 ⫾ (sd) 6.6 yr hirsutism or hyperandrogenemia and menstrual dysfunction (n ⫽ 21)
with a body mass index (BMI) 29.0 ⫾ (sd) 8.4 kg/m2, were studied. were diagnosed with PCOS according to the criteria derived from the
Hirsutism, as defined by the presence of excessive body hair distributed National Institute for Child Health and Human Development 1990s
in an androgen-dependent pattern, with a modified Ferriman-Gallwey conference (2). Hirsute patients with increased circulating concentra-
score (26) of 8 or more, was present in 57 of the patients with a score of tions of T, FT, DHEAS, and/or ⌬4-A and regular menstrual cycles were
14.6 ⫾ (sd) 5.3. considered to have hyperandrogenemic hirsutism (n ⫽ 28). Hirsute
Menstrual cycle intervals were evaluated on recall for every patient. patients with normal serum T, FT, DHEAS, and ⌬4-A levels and regular
Oligomenorrhea was defined by the presence of 6 or more cycles of more menstrual cycles were diagnosed with idiopathic hirsutism (n ⫽ 11).
than 36 days in the previous year, and amenorrhea by lack of vaginal Independently, the presence or absence of functional ovarian hyperan-
bleeding for 3 consecutive months (27). Oligomenorrhea and amenor- drogenism (FOH), as defined by increased 17-OHP levels after leupro-
rhea were considered as indicative of ovulatory dysfunction, but no lide administration, was also considered.
further effort was made to demonstrate oligo-ovulation. Oligomenor- The 95th percentile upper limits of normality for basal serum andro-
rhea or amenorrhea were present in 21 of the 60 patients, including 18 gens, derived from the control group of 27 healthy women, were 2.15

Downloaded from [Link] by guest on 07 May 2023

of the 57 hirsute patients and 3 women who did not have hirsutism (their nmol/liter, 35 pmol/liter, 9.5 ␮mol/liter, and 15.7 nmol/liter for T, FT,
hirsutism scores were 6, 7, and 7) but presented with oligomenorrhea DHEAS, and ⌬4-A, respectively. The 95th percentile upper limit of
and increased serum total T levels. normality for the leuprolide test, derived from 19 women in the control
The reference values for the analytical procedures were obtained from group, was a serum 17-OHP level of 7.6 nmol/liter.
a control group of 27 normal menstruating women, aged 30.2 ⫾ (sd) 8.7
yr, matched for BMI (28.9 ⫾ (sd) 7.7 kg/m2), who did not have signs and DNA extraction and genotype analysis
symptoms of hyperandrogenism or family history of endocrine diseases.
None of the patients had hypertension, features of Cushing’s disease, Genomic DNA was extracted from leukocytes obtained from whole
or drug-induced hirsutism. Hyperprolactinemia and congenital adrenal blood samples, using commercial DNA purification kits (Wizard
hyperplasia were ruled out because all the patients presented basal Genomic DNA purification kit, Promega Corp., Madison, WI, and Nu-
serum PRL levels ⬍24 ␮g/liter, together with ACTH-stimulated 17- cleon BAC C3, Amersham Pharmacia Biotech, Buckinghamshire, United
hydroxyprogesterone (17-OHP) levels ⬍30 nmol/liter and ACTH-stim- Kingdom). PCR primers were designed to amplify 2 fragments of the
ulated 11-deoxycortisol levels (S) ⬍26 nmol/liter (which is the mean ⫹ promoter of TNF␣ gene. The first fragment, spanning from ⫺1232 to
2 sd of the values obtained from the control group). ⫺732 relative to the TNF␣ gene transcription start site, comprises a DNA
Data from some patients and controls, regarding different aspects of region that contains the ⫺1196C/T, ⫺1125G/C, ⫺1031T/C, ⫺863C/A,
the pathophysiology of hirsutism, have been previously published (28). and ⫺857C/T polymorphisms described previously (35). Primer se-
The patients and controls had not taken hormonal medications, includ- quences were 5⬘-TCT GCT TGT GTG TGT GTG TCT G-3⬘ (sense) and
ing contraceptive pills, for the last 6 months. All the patients and controls 5⬘-ATG AAG CTC TCA CTT CTC AGG G-3⬘ (antisense). A second
were Caucasian. The ethics committee of the Hospital Ramón y Cajal fragment, spanning from ⫺444 to ⫺88 and containing the ⫺316G/A,
approved the study, and informed consent was obtained from each ⫺308G/A, ⫺238G/A, and ⫺163G/A polymorphisms (36), was also
patient and control. amplified. Primer sequences were 5⬘-CAA CGG ACT CAG CTT TCT
GAA G-3⬘ (sense) and 5⬘-TGG AGA AAC CCA TGA GCT CAT C-3⬘
(antisense).
Study protocol, hormone profiles, and diagnostic categories DNA sequences were amplified in an automated thermocycler
of hyperandrogenism (GeneAmp PCR System 2400, PE Applied Biosystems, Foster City, CA),
using a 25-␮l reaction mix containing 1 U of AmpliTaq polymerase (PE
Studies were performed between days 5 and 10 of the menstrual cycle Applied Biosystems). PCR conditions for both fragments included an
or during amenorrhea, after excluding pregnancy by proper testing. The initial denaturating step at 94 C for 1 min, 30 cycles at 94 C for 1 min,
patients reported to the Endocrine-Metabolic testing room between 0800 at 60 C for 1 min, and at 72 C for 1 min, and a final extension at 72 C
and 0900 h after a 12-h overnight fast. An indwelling iv line was placed for 10 min.
in a forearm vein, and after 15–30 min, basal blood samples were ob- After PCR amplification, alleles for each polymorphism were iden-
tained for the measurement of TNF␣, total T, ⌬4-androstenedione (⌬4-A), tified by direct sequence as follows. PCR products were purified using
17-OHP, dehydroepiandrosterone-sulfate (DHEAS), cortisol (F), LH, QIAquick purification kit (QIAGEN GmbH, Hilden, Germany). Se-
FSH, E2, SHBG, glucose, and insulin. Immediately after taking basal quence analysis was carried out using the dRhodamine Terminator
samples, a 250-␮g iv bolus of 1–24 ACTH (Synacthen, Ciba-Geigy, Basel, Cycle DNA Sequencing Ready Reaction Kit (PE Biosystems, Warrinton,
Switzerland) was injected and blood samples were obtained at 0 and 60 UK) using the antisense primer for the first fragment of the TNF␣ gene
min for the measurement of S, 17-OHP, and ⌬4-A. promoter and the sense primer for the second fragment. The cycle-
Finally, in all the patients and 19 controls, a 10-␮g/kg body weight sequencing products were precipitated and analyzed by an ABI 310
sc dose of leuprolide was injected (29), and samples were obtained at automated sequencer (PE Applied Biosystems) according to the man-
0900 h on the next day for measurement of LH, FSH, E2, F, T, SHBG, ufacturer’s instructions.
17-OHP, and ⌬4-A. Serum LH, FSH, and E2 served to confirm the
stimulatory effect of leuprolide on gonadotropin and ovarian steroid Statistical analysis
secretion, and F levels were measured in all the subjects to rule out an
interference of the stimulatory effect of the ACTH test performed the day Results are expressed as mean [median] ⫾ sd in the text and tables,
before, on serum 17-OHP and ⌬4-A levels. Samples were immediately unless otherwise stated. The Kolmogorov-Smirnov statistic, with a Lil-
centrifuged and serum was separated and frozen at ⫺20 C until assayed. liefors significance level for testing normality, was applied to continuous
Serum TNF␣ levels were measured within a single assay by solid- variables. Because most of the variables did not follow a normal dis-
phase, two-site chemiluminescent enzyme immunometric technique tribution, nonparametric tests were applied. The Mann-Whitney U -
(Immulite, Diagnostic Products Corp., Los Angeles, CA) with a sensi- Wilcoxon rank sum W test or the Kruskall-Wallis test were used to
tivity of 1.7 pg/ml and a mean intra-assay coefficient of variation of compare variables among the groups of subjects, as appropriate. For
3.2%. The technical characteristics of the assays employed for hormone variables showing significant differences between the groups by the
measurements have been reported previously (30, 31). The free testos- Kruskall-Wallis test, repeated Mann-Whitney U -Wilcoxon rank sum W
terone (FT) concentration was calculated from T and SHBG concentra- tests were used to identify the differences between each pair of groups,
tions, assuming a serum albumin concentration of 43 g/liter, and taking applying an a priori downward correction to the level of significance to
a value of 1 ⫻ 109 liter/mol for the association constant of SHBG for T compensate for multiple comparisons (i.e. P ⬍ 0.0125 is needed to reach
and a value of 3.6 ⫻ 104 liter/mol for that of albumin for T (32). statistically significant differences when comparisons involve four
Insulin resistance was estimated from fasting glucose and insulin groups) (37). A ␹2 test was used for discontinuous variables, and Spear-
Escobar-Morreale et al. • TNF-␣ and Hyperandrogenism The Journal of Clinical Endocrinology & Metabolism, August 2001, 86(8):3761–3767 3763

man’s nonparametric correlation analysis was also used. P ⬍ 0.05 was group with hyperandrogenic hirsutism) were diagnosed
considered statistically significant, with the exception stated above. with FOH (40) according to increased 17-OHP levels after the
Power analysis was performed using the G•Power software (38). Be-
cause power analysis requires parametric tests instead of the nonpara-
leuprolide challenge (␹2 ⫽ 3.936, P ⬍ 0.05 by Fisher’s exact
metric tests used here, sample sizes for 0.80 statistical power were test for the association between FOH and PCOS in patients).
calculated for the equivalent parametric tests (i.e., t test or one-way Serum TNF␣ levels were increased, compared with controls
ANOVA), and the results were corrected for the asymptotic relative only in the group of non-FOH patients, and FOH patients
efficiency of the nonparametric tests relative to their parametric equiv- showed intermediate values that were not different with
alents (39).
respect to those of controls and non-FOH patients (controls,
Results 5.6[4.4] ⫾ 4.0 pg/ml; FOH, 6.1[4.9] ⫾ 3.2 pg/ml, non-FOH,
7.6[7.2] ⫾ 3.3 pg/ml, P ⬍ 0.05). No differences were observed
Serum TNF␣ levels and hormone profiles
in HOMAIR or QUICKI indexes of insulin resistance or in the
When considered as a whole, the group of 60 hyperan- BMI when considering the diagnostic categories of hyperan-
drogenic patients presented with mildly increased serum drogenism or the presence or absence of FOH (data not
TNF␣ levels as compared with the healthy controls, although

Downloaded from [Link] by guest on 07 May 2023

shown).
there was a considerable overlap among these groups
(Fig. 1). Single nucleotide polymorphisms in the TNF␣ gene
Patients presented increased serum T, FT, basal and
ACTH-stimulated ⌬4-A and 17-OHP, and DHEAS concen- None of the polymorphisms in the promoter of the TNF␣
trations, and decreased SHBG levels, as compared with con- gene studied here was associated with patient or control
trols (Table 1). No differences were observed in basal LH, status (Table 2), with the diagnostic category of hyperan-
FSH, E2, and insulin resistance as measured by HOMAIR drogenism (data not shown) or with the presence or absence
and QUICKI indexes (Table 1). Serum TNF␣ levels showed of FOH (data not shown).
weak but statistically significant correlations, with basal and The frequencies of these polymorphisms in hyperandro-
ACTH-stimulated serum ⌬4-A concentrations (r ⫽ 0.286, P ⬍ genic patients and healthy controls are summarized in Table
0.01; and r ⫽ 0.222, P ⬍ 0.05, respectively), when considering 2. However, because of the limited sample size, our study
patients and controls as a whole. No other statistically sig- does not have enough statistical power to detect true but
nificant correlations were observed (data not shown). small differences in the proportions of affected and nonaf-
When considering the clinical diagnostic categories of hy- fected individuals among patients and controls (Table 2).
perandrogenism, the differences in serum TNF␣ levels Nevertheless, none of the subjects presented the ⫺163A and
among controls and patients with PCOS (n ⫽ 21), hyperan- the ⫺1196T variants, and only one patient presented the
drogenemic hirsutism (n ⫽ 28), and idiopathic hirsutism (n ⫽ ⫺1125C variant.
11) did not reach statistical significance (controls, 5.6[4.4] ⫾ We then studied the influence of these polymorphisms on
4.0 pg/ml; PCOS, 7.3[6.7] ⫾ 3.7 pg/ml; hyperandrogenemic the serum TNF␣ levels and on the hormonal profiles. When
hirsutism 6.8[7.0] ⫾ 2.7 pg/ml; and idiopathic hirsutism, considering hyperandrogenic patients and controls as a
8.1[7.6] ⫾ 4.1 pg/ml; P ⫽ 0.099). However, the statistical whole, carriers of the ⫺308A variant presented with in-
power was not enough to rule out such a difference (a total creased basal FT, 17-OHP, ⌬4-A, and DHEAS levels and
sample size of 209 subjects is needed for 0.80 power). increased serum T, FT, 17-OHP, and ⌬4-A levels after leu-
Fourteen patients (8 from the PCOS group and 6 from the prolide administration, compared with subjects with wild-
type (⫺308G) alleles (Table 3). No differences were observed
in serum TNF␣, basal T, ACTH-stimulated 17-OHP, and
⌬4-A and basal and leuprolide-stimulated LH, FSH, and E2
levels or in insulin resistance as measured by HOMAIR and
QUICKI indexes (Table 3). All these parameters showed no
statistically significant differences when studying patients
and controls separately.
Also, no differences were observed in serum determina-
tions among individuals with wild-type alleles, compared
with subjects hetero- or homozygous for the ⫺238A, ⫺316A,
⫺857T, ⫺863A, and ⫺1031C variants (data not shown).

Influence of obesity in serum TNF␣ levels and association

with TNF␣ gene polymorphisms
As stated above, the BMI was used to match patients and
controls, explaining why the study groups showed no sta-
FIG. 1. Serum concentrations of TNF␣ in hyperandrogenic patients tistically significant differences for this variable. For com-
(n ⫽ 60), compared with healthy controls (n ⫽ 27). The dot scatter- parisons between lean and obese subjects (obesity was de-
gram shows the individual data, whereas the box plot includes the fined by a BMI ⱖ 25 kg/m2), patients and controls were
median (horizontal line) and the interquartile range (box), and the
whiskers indicate the 5th and 95th percentiles. The groups were studied as a whole. Obese subjects presented higher serum
different with P ⬍ 0.02, by the Mann-Whitney U-Wilcoxon rank sum TNF␣ concentrations (7.4[7.1] ⫾ 4.0 vs. 5.6[5.0] ⫾ 2.6 pg/ml,
W test. P ⬍ 0.05), higher basal and leuprolide-stimulated FT levels
3764 The Journal of Clinical Endocrinology & Metabolism, August 2001, 86(8):3761–3767 Escobar-Morreale et al. • TNF-␣ and Hyperandrogenism

TABLE 1. Serum basal, ACTH-stimulated, and leuprolide-stimulated serum hormone concentrations and indexes of insulin resistance in
hyperandrogenic patients and healthy controls

Patients Controls
Variable P
(n ⫽ 60) (n ⫽ 27)
Basal T (nmol/liter) 2.2[2.1] ⫾ 0.8 1.4[1.5] ⫾ 0.4 ⬍0.001
Leuprolide-stimulated T (nmol/liter)a 2.2[2.0] ⫾ 0.9 1.6[1.7] ⫾ 0.6 ⬍0.01
Basal FT (pmol/liter) 40[38] ⫾ 19 18[17] ⫾ 8 ⬍0.001
Leuprolide-stimulated FT (pmol/liter)a 42[40] ⫾ 21 22[21] ⫾ 11 ⬍0.001
SHBG (nmol/liter) 38[31] ⫾ 23 68[66] ⫾ 31 ⬍0.001
Basal 17-OHP (nmol/liter) 3.4[2.8] ⫾ 2.3 2.3[2.0] ⫾ 1.2 ⬍0.02
ACTH-stimulated 17-OHP (nmol/liter) 9.5[8.1] ⫾ 5.4 7.1[6.4] ⫾ 3.5 ⬍0.005
Leuprolide-stimulated 17-OHP (nmol/liter)a 6.1[4.6] ⫾ 4.9 4.5[3.5] ⫾ 2.0 0.058
Basal ⌬4-A (nmol/liter) 13.1[4.9] ⫾ 2.3 9.6[9.5] ⫾ 3.2 ⬍0.001
ACTH-stimulated ⌬4-A (nmol/liter) 16.2[15.6] ⫾ 5.3 12.1[12.1] ⫾ 4.0 ⬍0.001
Leuprolide-stimulated ⌬4-A (nmol/liter)a 14.0[13.1] ⫾ 5.5 10.1[10.0] ⫾ 4.1 ⬍0.01

Downloaded from [Link] by guest on 07 May 2023

DHEAS (␮mol/liter) 7.6[6.7] ⫾ 4.3 5.3[5.3] ⫾ 2.4 ⬍0.01
Basal LH (IU/liter) 5.1[4.2] ⫾ 3.9 5.5[4.4] ⫾ 4.1 0.528
Leuprolide-stimulated LH (IU/liter)a 29.0[26.0] ⫾ 16.5 32.5[28.2] ⫾ 20.9 0.562
Basal FSH (IU/liter) 5.6[5.1] ⫾ 2.4 5.3[5.5] ⫾ 2.3 0.598
Leuprolide-stimulated FSH (IU/liter)a 12.4[11.4] ⫾ 5.5 14.4[14.6] ⫾ 4.9 0.090
Basal E2 (pmol/liter) 198[142] ⫾ 142 266[214] ⫾ 197 0.126
Leuprolide-stimulated E2 (pmol/liter)a 615[536] ⫾ 296 550[507] ⫾ 268 0.528
HOMAIR 3.1[2.7] ⫾ 1.5 2.6[2.4] ⫾ 1.7 0.163
QUICKI 0.33[0.33] ⫾ 0.02 0.35[0.34] ⫾ 0.07 0.163
a
The leuprolide test was performed in 19 of the 27 controls.

TABLE 2. Frequencies of the different single nucleotide polymorphisms (SNP) in the promoter of the TNF␣ gene in hyperandrogenic
patients and healthy controls

Observed frequencies
Actual Sample size needed to
SNP Controls Patients ␹2 P
sample size reach 0.80 power
% (affected/total) % (affected/total)
⫺163A 0.0 (0/24) 0.0 (0/57) 81
⫺238A 16.6 (4/24) 15.8 (9/57) 0.097 0.921 81 8780
⫺308A 20.8 (5/24) 29.8 (17/57) 0.690 0.406 81 156
⫺316A 12.5 (3/24) 15.8 (9/57) 0.145 0.704 81 570
⫺857T 29.2 (7/24) 38.5 (20/52)a 0.940 0.624 76 230
⫺863A 37.5 (9/24)b 32.7 (17/52)c 0.252 0.882 76 779
⫺1031C 54.1 (13/24)d 44.0 (22/50)e) 0.673 0.710 74 201
⫺1125C 0.0 (0/24) 2.1 (1/48) 0.507 0.476 72
⫺1196T 0.0 (0/24) 0.0 (0/47) 71
No statistically significant associations were observed when considering hetero- and homozygous subjects separately.
a
One patient was homozygous for this variant.
b
Two controls were homozygous for this variant.
c
Three patients were homozygous for this variant.
d
Three controls were homozygous for this variant.
e
Five patients were homozygous for this variant.

(basal: 38[35] ⫾ 22 vs. 25[23] ⫾ 13 pmol/liter, P ⬍ 0.010; 380.5, P ⫽ 0.105). However, the statistical power of the latter
leuprolide-stimulated: 42[41] ⫾ 23 vs. 29[27] ⫾12 pmol/liter, comparison was not enough to rule out such a difference (a
P ⬍ 0.010), lower SHBG levels (39[30] ⫾ 28 vs. 60[57] ⫾ 29 total sample size of 216 subjects is needed for 0.80 power).
nmol/liter, P ⬍ 0.010), higher HOMAIR (3.4[3.4] ⫾ 1.6 vs. Finally, none of the polymorphisms in the promoter of the
2.1[2.0] ⫾ 1.1, P ⬍ 0.001), and lower QUICKI (0.32[0.32] ⫾ TNF␣ gene were associated with obesity (data not shown).
0.02 vs. 0.36[0.35] ⫾ 0.06, P ⬍ 0.001), compared with lean
individuals. No differences were observed in other variables.
Discussion
These results persisted when studying patients alone, but
when the analysis was restricted to healthy controls, only the The possible involvement of TNF␣ in the pathogenesis of
differences in HOMAIR and QUICKI remained statistically hyperandrogenism is based on several recent findings. TNF␣
significant (data not shown). might be related to increased ovarian steroid secretion,
We then compared serum TNF␣ levels between patients anovulation, and ovarian apoptosis in animals (5–7), features
and controls in obese and in lean subjects separately. Serum that resemble those of hyperandrogenism in humans.
TNF␣ levels were higher in the 23 lean patients, compared Also, as reviewed by Hotamisligil (41), virtually all animal
with the 10 lean controls (6.2[5.1] ⫾ 2.2 vs. 4.3[3.1] ⫾ 3.0; U ⫽ and human models of obesity and insulin resistance are
63; W ⫽ 15; P ⬍ 0.05). On the contrary, serum TNF␣ levels associated with TNF␣ messenger RNA and protein hyper-
were not different between the 37 obese patients and the 17 expression. As stated above, obesity and insulin resistance
obese controls (7.8[7.9] ⫾ 3.8 vs. 6.4[6.5] ⫾ 4.5; U ⫽ 227.5; W ⫽ are frequent findings in hyperandrogenic women (42).
Escobar-Morreale et al. • TNF-␣ and Hyperandrogenism The Journal of Clinical Endocrinology & Metabolism, August 2001, 86(8):3761–3767 3765

TABLE 3. Serum TNF␣ levels and basal, ACTH-stimulated, and leuprolide-stimulated serum hormone concentrations in carriers of the
⫺308A variant, compared with subjects with wild-type alleles (⫺308G), considering patients and controls as a whole

⫺308A ⫺308G
Variable P
(n ⫽ 22) (n ⫽ 59)
TNF␣ (pg/ml) 6.9[6.5] ⫾ 4.2 6.8[6.5] ⫾ 3.5 0.992
Basal T (nmol/liter) 2.3[2.0] ⫾ 0.8 1.9[1.7] ⫾ 0.8 0.051
Leuprolide-stimulated T (nmol/liter)a 2.5[2.5] ⫾ 1.1 2.0[1.9] ⫾ 0.8 ⬍0.05b
Basal FT (pmol/liter) 42[37] ⫾ 21 32[28] ⫾ 19 ⬍0.05b
Leuprolide-stimulated FT (pmol/liter)a 48[46] ⫾ 21 36[33] ⫾ 19 ⬍0.05b
SHBG (nmol/liter) 37[30] ⫾ 19 47[40] ⫾ 28 0.152
Basal 17-OHP (nmol/liter) 4.0[3.6] ⫾ 2.6 2.8[2.5] ⫾ 1.8 ⬍0.05b
ACTH-stimulated 17-OHP (nmol/liter) 9.5[8.4] ⫾ 4.6 8.8[7.4] ⫾ 5.3 0.179
Leuprolide-stimulated 17-OHP (nmol/liter)a 6.8[5.8] ⫾ 3.2 5.4[4.1] ⫾ 4.8 ⬍0.05
Basal ⌬4-A (nmol/liter) 14.1[13.3] ⫾ 4.5 11.8[11.8] ⫾ 4.6 ⬍0.05b
ACTH-stimulated ⌬4-A (nmol/liter) 15.9[13.8] ⫾ 5.1 15.1[15.2] ⫾ 5.1 0.698

Downloaded from [Link] by guest on 07 May 2023

Leuprolide-stimulated ⌬4-A (nmol/liter)a 15.6[14.4] ⫾ 5.3 12.7[11.9] ⫾ 5.1 ⬍0.05b
DHEAS level (␮mol/liter) 8.6[7.7] ⫾ 5.1 6.4[5.9] ⫾ 3.4 ⬍0.05b
Basal LH (IU/liter) 5.8[4.6] ⫾ 4.1 5.1[4.2] ⫾ 3.8 0.414
Leuprolide-stimulated LH (IU/liter)a 35.8[30.9] ⫾ 20.6 27.9[26.7] ⫾ 15.3 0.164
Basal FSH (IU/liter) 5.2[5.1] ⫾ 1.5 5.7[5.0] ⫾ 2.6 0.803
Leuprolide-stimulated FSH (IU/liter)a 13.5[12.6] ⫾ 5.2 12.6[11.3] ⫾ 5.5 0.325
Basal E2 (pmol/liter) 224[192] ⫾ 139 222[142] ⫾ 174 0.342
Leuprolide-stimulated E2 (pmol/liter)a 628[525] ⫾ 258 617[532] ⫾ 301 1.000
HOMAIR 3.1[3.3] ⫾ 1.4 2.9[2.5] ⫾ 1.7 0.284
QUICKI 0.33[0.32] ⫾ 0.02 0.34[0.33] ⫾ 0.05 0.284
BMI (kg/m2) 30.1[29.8] ⫾ 6.5 28.9[26.3] ⫾ 8.9 0.222
a
Because the leuprolide test was performed in only 19 controls, these comparisons include 20 carriers of the ⫺308A variant and 53 subjects
with ⫺308G alleles.
b
Although no statistically significant differences in BMI, HOMAIR, and QUICKI were observed, analysis of covariance was performed to
rule out a possible influence of these variables on the differences found in serum hormone levels. Normality was ensured by logarithmic
transformation before analysis of covariance. After controlling for BMI, HOMAIR, and QUICKI, leuprolide-stimulated T, basal and leuprolide-
stimulated FT, basal 17-OHP, and basal and leuprolide-stimulated ⌬4-A serum concentrations remained significantly increased in carriers of
the ⫺308A variant as compared with subjects homozygous for the ⫺308G allele.

Our present results suggest that TNF␣ might play a role acts mostly if not completely by autocrine or paracrine mech-
in the etiology of hyperandrogenism. Serum TNF␣ levels anisms (43), it is also possible that the fluctuations in serum
were increased, although with significant overlap, in pa- TNF␣ levels may have no pathogenic significance, repre-
tients, compared with healthy controls. Because both groups senting only secondary events not actually related to the
were matched for BMI, this increase is apparently indepen- development of hyperandrogenism.
dent of obesity. Moreover, in our series the indexes of insulin Nevertheless, in our series the TNF␣ system also influ-
resistance were similar in patients and controls. Therefore, enced hyperandrogenism from a genetic perspective. Al-
the mildly increased serum TNF␣ levels found in our hy- though none of the polymorphisms studied here was more
perandrogenic patients appear to be also independent of frequent in hyperandrogenic patients, the ⫺308A variant in
insulin resistance. the promoter of the TNF␣ gene clearly influenced the phe-
Obesity alone modulates serum TNF␣ levels, which, as has notype, resulting in increased basal and leuprolide-stimu-
been described by others (16, 17), were higher in obese sub- lated androgen concentrations in the group of carriers of this
jects, compared with lean individuals. Obese subjects also variant.
presented a higher degree of insulin resistance, lower SHBG, The ⫺308G/A polymorphism in the promoter of the TNF␣
and increased FT, compared with lean subjects. When sub- gene has been studied extensively. The ⫺308A variant,
jects were classified by body weight, serum TNF␣ levels were which is associated with human leukocyte antigens A1, B8,
increased only in lean patients, compared with lean controls, and DR3 alleles (44), is a much more powerful transcription
but this difference was not maintained when comparing activator, compared with the ⫺308G allele (45, 46), explain-
obese patients with obese controls. Gonzalez et al. (8) recently ing the increased TNF␣ production found in these individ-
reported similar results in PCOS patients. uals (47).
Therefore, our present results confirm that serum TNF␣ Based on these previous studies, we hypothesize that car-
levels increase mainly because of obesity both in controls and riers of the ⫺308A allele might have increased TNF␣ pro-
hyperandrogenic women but also point to a different mech- duction in several tissues, including the ovary. In such a case,
anism in relation to hyperandrogenism. Interestingly, the the increased TNF␣ levels may stimulate ovarian ⌬4-A se-
higher serum TNF␣ levels were found in the less severe cretion in theca cells as occurs in experimental animals (5).
hyperandrogenic women according to the leuprolide test, However, definite proof would require in vitro studies that
and serum TNF␣ levels tended to be higher in women with are far beyond the methodology of our present study.
idiopathic hirsutism. Because the increased androgen secretion in ⫺308A car-
Whether these tendencies reflect a pathogenic mechanism riers was observed when considering patients and controls
cannot be solved by our present results but, because TNF␣ as a whole, the ⫺308A variant in the promoter of the TNF␣
3766 The Journal of Clinical Endocrinology & Metabolism, August 2001, 86(8):3761–3767 Escobar-Morreale et al. • TNF-␣ and Hyperandrogenism

gene should be considered only as a contributing factor to the Acknowledgments

development of hyperandrogenism instead of the main eti- We thank Ms. Genoveva González for technical assistance.
ologic factor for this condition. Interestingly, serum TNF␣
levels were similar in ⫺308A carriers and in subjects pre- Received December 5, 2000. Accepted April 20, 2001.
senting wild-type alleles, pointing to a local effect of TNF␣ Address all correspondence and requests for reprints to: Héctor F.
on androgen secreting tissue rather than to an endocrine Escobar-Morreale, M.D., Ph.D., Department of Endocrinology, Hospital
Ramón y Cajal, Carretera de Colmenar km. 9,100, 28034 Madrid, Spain.
effect mediated by circulating TNF␣. E-mail: [Link]@[Link].
On the contrary, neither the ⫺308G/A polymorphism nor This work was supported by a grant (Proyecto 08.6/0022/1998 to
any of the other TNF␣ gene polymorphisms studied here H.F.E.-M.) from the Consejerı́a de Investigación y Cultura, Comunidad
were associated with patient or control status. Milner et al. de Madrid, Spain, and by grants (Proyecto FIS 00/0414 to H.F.E.-M.;
(48) recently reported similar results, failing to demonstrate Contrato de Investigador FIS 98/3044 to R.M.C.) from the Fondo de
Investigación Sanitaria, Ministerio de Sanidad y Consumo, Spain. This
an association of the ⫺308G/A polymorphism with PCOS in work was presented at the 82nd Annual Meeting of The Endocrine
their series. However, Milner et al. (48) did not observe dif- Society, Toronto, Canada, June 21–24, 2000.

Downloaded from [Link] by guest on 07 May 2023

ferences in serum androgens depending on the ⫺308G/A
polymorphism, but they measured only T and ⌬4-A.
The association of variants in the promoter of the TNF␣ References
gene with obesity and insulin resistance is controversial, and 1. Asunción M, Calvo RM, San Millán JL, Sancho J, Avila S, Escobar-Morreale
HF 2000 A prospective study of the prevalence of the polycystic ovary syn-
there have been negative (36, 49 –52) and positive (18 –20) drome in unselected Caucasian women from Spain. J Clin Endocrinol Metab
reports for these associations over the last years. In our series, 85:2434 –2438
none of the polymorphisms in the promoter of the TNF␣ 2. Zawadzki JK, Dunaif A 1992 Diagnostic criteria for polycystic ovary syn-
drome: towards a rational approach. In: Dunaif A, Givens JR, Haseltine FP,
gene were associated with obesity, and these polymorphisms Merriam GR, eds. Polycystic ovary syndrome. Boston: Blackwell Scientific
did not influence serum TNF␣ levels, BMI, HOMAIR, or Publications; 377–384
QUICKI, which were not different among the carriers of 3. Legro RS, Driscoll D, Strauss 3rd JF, Fox J, Dunaif A 1998 Evidence for a
genetic basis for hyperandrogenemia in polycystic ovary syndrome. Proc Natl
these variants and subjects presenting wild-type alleles. Al- Acad Sci USA 95:14956 –14960
beit we have assessed insulin resistance by relatively insen- 4. Legro RS, Spielman R, Urbanek M, Driscoll D, Strauss 3rd JF, Dunaif A 1998
Phenotype and genotype in polycystic ovary syndrome. Recent Prog Horm Res
sitive methods, our present results cannot demonstrate any 53:217–256
association between obesity and insulin resistance with the 5. Roby KF, Terranova PF 1990 Effects of tumor necrosis factor-␣ in vitro on
polymorphisms in the promoter of the TNF␣ gene studied steroidogenesis of healthy and atretic follicles of the rat: theca as a target.
Endocrinology 126:2711–2718
here. 6. Spaczynski RZ, Arici A, Duleba AJ 1999 Tumor necrosis factor-␣ stimulates
Sample size and statistical power merits an explanation. proliferation of rat ovarian theca-interstitial cells. Biol Reprod 61:993–998
Statistical power, as the complement of type II error, reflects 7. Kaipia A, Chun SY, Eisenhauer K, Hsueh AJ 1996 Tumor necrosis factor-␣
and its second messenger, ceramide, stimulate apoptosis in cultured ovarian
the ability of a study to detect a true difference (37). There- follicles. Endocrinology 137:4864 – 4870
fore, low power indicates an elevated probability of con- 8. Gonzalez F, Thusu K, Abdel-Rahman E, Prabhala A, Tomani M, Dandona
cluding erroneously that there are no differences or associ- P 1999 Elevated serum levels of tumor necrosis factor ␣ in normal-weight
women with polycystic ovary syndrome. Metabolism 48:437– 441
ations in a study. As a rule, a 0.80 power is considered 9. Jasper M, Norman RJ 1995 Immunoactive interleukin-1 beta and tumour
adequate (37). Our present study lacks enough power to necrosis factor-␣ in thecal, stromal and granulosa cell cultures from normal and
polycystic ovaries. Hum Reprod 10:1352–1354
accurately rule out that the proportions of affected and non- 10. Hotamisligil GS, Arner P, Caro JF, Atkinson RL, Spiegelman BM 1995
affected individuals for the polymorphisms in the TNF␣ Increased adipose tissue expression of tumor necrosis factor-␣ in human obe-
gene were actually different among hyperandrogenic pa- sity and insulin resistance. J Clin Invest 95:2409 –2415
11. Saghizadeh M, Ong JM, Garvey WT, Henry RR, Kern PA 1996 The expression
tients and controls. of TNF ␣ by human muscle. Relationship to insulin resistance. J Clin Invest
Yet even in the case that these differences actually exist, 97:1111–1116
they are too small (the higher was a 10-point difference in the 12. Hotamisligil GS, Peraldi P, Budavari A, Ellis R, White MF, Spiegelman BM
1996 IRS-1-mediated inhibition of insulin receptor tyrosine kinase activity in
percentage of affected individuals among patients and con- TNF-␣- and obesity-induced insulin resistance. Science 271:665– 668
trols) to represent a major mechanism in the pathogenesis of 13. Hotamisligil GS, Arner P, Atkinson RL, Spiegelman BM 1997 Differential
regulation of the p80 tumor necrosis factor receptor in human obesity and
hyperandrogenism. Moreover, as stated above, the study by insulin resistance. Diabetes 46:451– 455
Milner et al. (48) ruled out an association of the ⫺308A variant 14. Hube F, Birgel M, Lee YM, Hauner H 1999 Expression pattern of tumour
with PCOS, and the sample size in this study provided ad- necrosis factor receptors in subcutaneous and omental human adipose tissue:
role of obesity and non-insulin-dependent diabetes mellitus. Eur J Clin Invest
equate statistical power. 29:672– 678
In conclusion, our present results suggest that the TNF␣ 15. Miles PD, Romeo OM, Higo K, Cohen A, Rafaat K, Olefsky JM 1997 TNF-
system might play a role in the pathogenesis of hyperan- ␣-induced insulin resistance in vivo and its prevention by troglitazone. Dia-
betes 46:1678 –1683
drogenism, independent of obesity and insulin resistance. 16. Katsuki A, Sumida Y, Murashima S, et al. 1998 Serum levels of tumor necrosis
Not only serum TNF␣ levels were increased in patients, factor-␣ are increased in obese patients with noninsulin-dependent diabetes
compared with controls, but also the ⫺308G/A polymor- mellitus. J Clin Endocrinol Metab 83:859 – 862
17. Dandona P, Weinstock R, Thusu K, Abdel-Rahman E, Aljada A, Wadden T
phism in the promoter of the TNF␣ gene modulated ovarian 1998 Tumor necrosis factor-␣ in sera of obese patients: fall with weight loss.
function, resulting in increased serum androgen levels in J Clin Endocrinol Metab 83:2907–2910
18. Fernandez-Real JM, Gutierrez C, Ricart W, et al. 1997 The TNF-␣ gene Nco
carriers of the ⫺308A variant. However, elucidation of the I polymorphism influences the relationship among insulin resistance, percent
precise mechanisms underlying the relationship between the body fat, and increased serum leptin levels. Diabetes 46:1468 –1472
TNF␣ system and androgen excess is needed before consid- 19. Herrmann SM, Ricard S, Nicaud V, et al. 1998 Polymorphisms of the tumour
necrosis factor-␣ gene, coronary heart disease and obesity. Eur J Clin Invest
ering TNF␣ as a significant contributing factor to the devel- 28:59 – 66
opment of hyperandrogenism. 20. Hoffstedt J, Eriksson P, Hellstrom L, Rossner S, Ryden M, Arner P 2000
Escobar-Morreale et al. • TNF-␣ and Hyperandrogenism The Journal of Clinical Endocrinology & Metabolism, August 2001, 86(8):3761–3767 3767

Excessive fat accumulation is associated with the TNF␣-308 G/A promoter 36. Hamann A, Mantzoros C, Vidal-Puig A, Flier JS 1995 Genetic variability in
polymorphism in women but not in men. Diabetologia 43:117–120 the TNF-␣ promoter is not associated with type II diabetes mellitus (NIDDM).
21. Conn JJ, Jacobs HS, Conway GS 2000 The prevalence of polycystic ovaries in Biochem Biophys Res Commun 211:833– 839
women with type 2 diabetes mellitus. Clin Endocrinol (Oxf) 52:81– 86 37. Dawson-Saunders B, Trapp RG 1990 Basic and clinical biostatistics. East
22. Legro RS, Kunselman AR, Dodson WC, Dunaif A 1999 Prevalence and Norwalk, CT: Appleton & Lange
predictors of risk for type 2 diabetes mellitus and impaired glucose tolerance 38. Buchner A, Faul F, Erdfelder E 1997 G䡠Power: a priori, post-hoc, and com-
in polycystic ovary syndrome: a prospective, controlled study in 254 affected promise power analyses for the Macintosh. Version 2.1.2. Trier, Germany:
women. J Clin Endocrinol Metab 84:165–169 University of Trier
23. Moghetti P, Tosi F, Castello R, et al. 1996 The insulin resistance in women with 39. Lehmann EL 1975 Nonparametrics: statistical methods based on ranks. San
hyperandrogenism is partially reversed by antiandrogen treatment— evidence Francisco: Holden-Bay, McGraw-Hill, Inc.
that androgens impair insulin action in women. J Clin Endocrinol Metab 40. Ehrmann DA, Barnes RB, Rosenfield RL 1995 Hyperandrogenism, hirsutism,
81:952–960 and the polycystic ovary syndrome. In: DeGroot LJ, ed. Endocrinology. Phil-
24. Solomon CG 1999 The epidemiology of polycystic ovary syndrome. Preva- adelphia: Saunders; 2093–2112
lence and associated disease risks. Endocrinol Metab Clin North Am 28: 41. Hotamisligil GS 1999 The role of TNF␣ and TNF receptors in obesity and
247–263 insulin resistance. J Intern Med 245:621– 625
25. Urbanek M, Legro RS, Driscoll DA, et al. 1999 Thirty-seven candidate genes 42. Dunaif A 1997 Insulin resistance and the polycystic ovary syndrome: mech-
for polycystic ovary syndrome: strongest evidence for linkage is with follista- anism and implications for pathogenesis. Endocr Rev 18:774 – 800
tin. Proc Natl Acad Sci USA 96:8573– 8578 43. Hube F, Hauner H 1999 The role of TNF-␣ in human adipose tissue: prevention

Downloaded from [Link] by guest on 07 May 2023

26. Hatch R, Rosenfield RL, Kim MH, Tredway D 1981 Hirsutism: implications, of weight gain at the expense of insulin resistance? Horm Metab Res 31:
etiology, and management. Am J Obstet Gynecol 140:815– 830 626 – 631
27. Goodman A 1999 Terminology and differential diagnosis of genital tract 44. Wilson AG, de Vries N, Pociot F, di Giovine FS, van der Putte LB, Duff GW
bleeding in women. In: Snyder PJ, Utiger RD, eds. Up To Date in Endocri-
1993 An allelic polymorphism within the human tumor necrosis factor ␣
nology and Diabetes. Wellesley, MA: UpToDate, Inc
promoter region is strongly associated with HLA A1, B8, and DR3 alleles. J Exp
28. Calvo RM, Asunción M, Sancho J, San Millán JL, Escobar-Morreale HF 2000
Med 177:557–560
The role of the CAG repeat polymorphism in the androgen-receptor gene and
45. Wilson AG, Symons JA, McDowell TL, McDevitt HO, Duff GW 1997 Effects
of skewed X-chromosome inactivation, in the pathogenesis of hirsutism. J Clin
of a polymorphism in the human tumor necrosis factor ␣ promoter on tran-
Endocrinol Metab 85:1735–1740
scriptional activation. Proc Natl Acad Sci USA 94:3195–3199
29. Ehrmann DA, Schneider DJ, Sobel BE, et al. 1997 Troglitazone improves
defects in insulin action, insulin secretion, ovarian steroidogenesis, and fibri- 46. Kroeger KM, Carville KS, Abraham LJ 1997 The ⫺308 tumor necrosis factor-␣
nolysis in women with polycystic ovary syndrome. J Clin Endocrinol Metab promoter polymorphism effects transcription. Mol Immunol 34:391–399
82:2108 –2116 47. Jacob CO, Fronek Z, Lewis GD, Koo M, Hansen JA, McDevitt HO 1990
30. Escobar-Morreale HF, Serrano-Gotarredona J, Varela C, Garcı́a-Robles R, Heritable major histocompatibility complex class II-associated differences in
Sancho JM 1997 Circulating leptin concentrations in women with hirsutism. production of tumor necrosis factor ␣: relevance to genetic predisposition to
Fertil Steril 68:898 –906 systemic lupus erythematosus. Proc Natl Acad Sci USA 87:1233–1237
31. Escobar-Morreale HF, Avila S, Sancho J 2000 Serum prostate-specific antigen 48. Milner CR, Craig JE, Hussey ND, Norman RJ 1999 No association between
concentrations are not useful for monitoring the treatment of hirsutism with the ⫺308 polymorphism in the tumour necrosis factor ␣ (TNF␣) promoter
oral contraceptive pills. J Clin Endocrinol Metab 85:2488 –2492 region and polycystic ovaries. Mol Hum Reprod 5:5–9
32. Vermeulen A, Verdonck L, Kaufman JM 1999 A critical evaluation of simple 49. Fernandez-Real JM, Gutierrez C, Ricart W, et al. 1997 The TNF-beta gene Nco
methods for the estimation of free testosterone in serum. J Clin Endocrinol I polymorphism is not associated with hypertriglyceridemia or insulin resis-
Metab 84:3666 –3672 tance in lean and obese subjects. Biochem Biophys Res Commun 236:829 – 832
33. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC 50. Walston J, Seibert M, Yen CJ, Cheskin LJ, Andersen RE 1999 Tumor necrosis
1985 Homeostasis model assessment: insulin resistance and ␤-cell function factor-␣⫺238 and ⫺308 polymorphisms do not associated with traits related
from fasting plasma insulin and glucose concentrations in man. Diabetologia to obesity and insulin resistance. Diabetes 48:2096 –2098
28:412– 419 51. Lee SC, Pu YB, Thomas GN, et al. 2000 Tumor necrosis factor ␣ gene G-308A
34. Katz A, Nambi SS, Mather K, et al. 2000 Quantitative insulin sensitivity check polymorphism in the metabolic syndrome. Metabolism 49:1021–1024
index: a simple, accurate method for assessing insulin sensitivity in humans. 52. Rasmussen SK, Urhammer SA, Jensen JN, Hansen T, Borch-Johnsen K,
J Clin Endocrinol Metab 85:2402–2410 Pedersen O 2000 The ⫺238 and ⫺308 G-⬎A polymorphisms of the tumor
35. Negoro K, Kinouchi Y, Hiwatashi N, et al. 1999 Crohn’s disease is associated necrosis factor ␣ gene promoter are not associated with features of the insulin
with novel polymorphisms in the 5⬘-flanking region of the tumor necrosis resistance syndrome or altered birth weight in Danish Caucasians. J Clin
factor gene. Gastroenterology 117:1062–1068 Endocrinol Metab 85:1731–1734